A Data-Driven Workload Prediction Model for Cloud Computing Using Machine Learning

Cloud computing (CC) technology is widely pay-per-use to provide efficient services to the user. Therefore, it is frequently used in business management because it is cheaper, scalable, and flexible but has some limitations. However, workload prediction is a challenging and complex problem. To address this issue, a data-driven workload prediction model inspired by a decision tree-based model of Machine Learning has been proposed using the benchmark dataset Google Cluster Workload Traces 2019. It is a large benchmark dataset used to analyze or predict workload behavior, resource allocation, or system performance. The main aim is to optimize user satisfaction and profit for cloud service providers and minimize the scalability of resources. The proposed model has been assessed using different evaluation parameters like RMSE, MSE, MAE, R-squared, training time, model size, and prediction speed. The results of this study show that the proposed model performs effectively and outperforms the contemporary model.